面向相关多敏感属性的隐私保护方法-计算机软件与理论专业毕业论文.docxVIP

I I 摘 要 摘 要 随着数据挖掘技术及网络技术的快速发展，数据发布中如何解决信息共享和隐私保 护问题成为了重要的研究内容。数据发布中常常会涉及到相关敏感属性，将现有的敏感 属性隐私保护方法直接应用于相关敏感属性的隐私保护会导致隐私数据泄漏。本文提出 了相关敏感属性的隐私保护方法。 本文在属性选择基础上提出了独立相关敏感属性和联合相关敏感属性的概念，并给 出相应的判别方法。通过该判别方法将相关敏感属性按其性质分类后进行隐私保护，并 讨论了隐私保护方法。本文中的隐私保护方法借鉴了有损连接对隐私数据进行保护的思 想，对关系表中的记录进行聚类，保证了关系表中记录按敏感等级划分；对已划分的记 录按照频率比较策略进行分组，提出了一种基于聚类前提下的(k, p)-sensitive相关敏感属 性数据分组算法。该方法的额外信息损失度较大，对其进行了改进。改进的(k, p)-sensitive 相关敏感属性数据分组算法采用循环聚类的方法处理剩余记录，增加记录分组，减少了 记录的二次添加。 实验结果表明(k, p)-sensitive算法可以有效防止隐私泄露，增强数据发布的安全性。 改进后的(k, p)-sensitive算法在有效防止隐私泄露的基础上，进一步减少了有损连接造成 的信息损失，并提高了数据的可共享性。 关键词 属性选择 数据隐私 相关敏感属性 多样性 独立相关 联合相关 I II Abstract Abstract With the rapid development of data mining technology and network technology, How to solve information sharing and privacy protection in the process of data publishing has become an important research. Data releasing is often involved relevant sensitive attributes in real applications. Directly applying the existing sensitive attribute privacy protection methods to multiple sensitive attributes privacy protection can divulge the privacy data. This paper proposes a method of protecting privacy which aims at data containing relevant multi-sensitive attribute. This paper proposes the concepts about independent relevant multi-sensitive attributes and joint relevant multi-sensitive attributes based on attribute selection .The records will be protected after classified by their nature, and discusses the method of privacy protection. The thought of lossy join for protecting privacy data is inherited, and the records are clustered to guarantee the sensitive rank division in tables. The records divided were grouped according to frequency of comparative strategy. And a grouping algorithm which is (k, p)-sensitive relevant multi-sensitive attribute method is proposed based on clustering. The experimental results indicated that this algorithm could prevent the privacy revelation, and strengthen security of data published. The method produces greater additional information